Browsing by Author "Laitonen, Janne"
Now showing 1 - 1 of 1
- Results Per Page
- Sort Options
- Todennäköisyyspohjainen riskien seuranta ydinvoimalaitosten valvonnassa
School of Science | Master's thesis(2010) Laitonen, JanneThis thesis examines risk follow-up in the framework of regulatory control of nuclear safety. In Finland, the power companies are obliged to report all operational events to the Radiation and Nuclear Safety Authority (STUK). First, the essential concepts of nuclear safety and probabilistic risk assessment (PRA) are introduced, for they build the foundation of risk follow-up. The thesis concentrates on level 1 PRA-models. In risk follow-up, the incremental conditional core damage probability of operational events is evaluated based on a plant specific PRA-model. The objective of this PRAbased event analysis is to classify the events based on their risk significance, follow the recurrence of the events during the life cycle of the nuclear power plant, and offer lessons learned for future improvements. The theoretical concepts and methodological problems of risk follow-up are introduced, one issue being retrospective probability assessment. This means that the probability of an accident is evaluated given the fact that no accident ever occurred. This problem is handled by separating the operational information into two modes. The first concerns initiating events and the second component failures. In 2009, no initiating events occurred at Finnish nuclear power plants. The most risk significant events at Loviisa site were recurrent failures in ventilation and at Olkiluoto site diesel generator failures. At STUK, risk follow-up is focused on the yearly risk-based classification of operational events. The number of these events varies every year depending on the assumed stochastic properties of the events. Therefore, it has been difficult to make reliable inferences on the abnormality of the yearly result. Attention has been paid to the absolute number of events and the significance of the stochastic process has not been evaluated. The method introduced in this thesis simulates component failures through a stochastic process and the risk significance is assessed based on a PRA-model. This way, an uncertainty distribution for the number of failures is obtained to help the inference and decision-making. As an example, events at Olkiluoto site are utilized. The results indicate that the parameters of the PRA-model are conservative, as expected.